Adaptive resource management system for home-area networks

We propose an adaptive resource management system that manages computational resources, such as disk space, memory and CPU power, in home-area networks comprised of various devices, such as personal computers and information appliances. This system enables "poor" devices with limited computational resources to borrow excess computational resources from "rich" devices. Existing distributed systems and software frameworks for home-area networks do not account for important characteristics of home-area network environments, such as user needs based on family members' interests and physical state changes of devices; e.g., disconnection from and reconnection to networks. The proposed system overcomes this problem by employing dynamic adaptation mechanisms. A prototype of this adaptive resource management system, called PPS/J, was designed and implemented on a Java virtual machine. The overhead of the dynamic adaptation mechanisms was measured on the PPS/J, and we found it sufficiently low during application program execution.